Class 707 - SWT: Introduction into service

[AM9]

more factual is that in this months Modern Railways the 707s are now firmly at the top of Roger Ford's new train reliability table with an amazingly high Miles Per Technical Incident of over 189,000, with a single TIN* in the reporting period. Some similar age fleets are still struggling to achieve 10,000

*any incident caused by rail vehicle fault resulting in a delay of 3 minutes or more

MTIN figures are OK for monitoring growth of availability and seasonal trends, but aren't they flawed when superficially comparing similar equipment running services with very different demands on their speeds, number of stops, loads track conditions etc.?

 

[jopsuk]

MTIN figures are OK for monitoring growth of availability and seasonal trends, but aren't they flawed when superficially comparing similar equipment running services with very different demands on their speeds, number of stops, loads track conditions etc.?

they are, but given that at the very least the other Desiro City fleets should at least be in the same ball park as the 707s, but the 717s and 700s are not even in the same county

 

[AM9]

they are, but given that at the very least the other Desiro City fleets should at least be in the same ball park as the 707s, but the 717s and 700s are not even in the same county

I can't speak for the class 717s and they are still quite new so they are still in availability growth, but the class 700s have a more complex operation and are worked quite hard (especially on ac) even though they have more power available than the 707s.

 

[Goldfish62]

I can't speak for the class 717s and they are still quite new so they are still in availability growth, but the class 700s have a more complex operation and are worked quite hard (especially on ac) even though they have more power available than the 707s.

700 MTIN is nowhere near showing the bathtub curve of improvement that it should be. 14K rolling average MTIN for an EMU that's been in service for a few years now is not up to expectations.

 

[hwl]

MTIN figures are OK for monitoring growth of availability and seasonal trends, but aren't they flawed when superficially comparing similar equipment running services with very different demands on their speeds, number of stops, loads track conditions etc.?

The number of cars per unit also has a big effect and whether they are join /split in service.
e.g. 8car 700 should be better than 12car 700 for example.
Any computer reset will take more than 3 minutes from the start of the problem so that could be contributing a lot. Getting software sorted doesn't really fit in with traditional MTIN type thinking.
With the 700s you have to wonder whether the comparative lack of maintenance time and staff is having a negative self reinforcing effect outside any software issues?

707s - that is good door reliability to achieve that. And not coupling units in service will also help.

 

[AM9]

The number of cars per unit also has a big effect and whether they are join /split in service.
e.g. 8car 700 should be better than 12car 700 for example.
Any computer reset will take more than 3 minutes from the start of the problem so that could be contributing a lot. Getting software sorted doesn't really fit in with traditional MTIN type thinking.
With the 700s you have to wonder whether the comparative lack of maintenance time and staff is having a negative self reinforcing effect outside any software issues?

707s - that is good door reliability to achieve that. And not coupling units in service will also help.

As an aside, is there an accessible online source of MTIN figures for all MUs? Many posters here quote them but maybe they are from sources available only to industry insiders. It would be useful to compare the oft quoted high figures of established three and four-car EMUs with the current growth of new 5/6/7/8/9 & 12-car units, particularly as the MTIN for example doesn't recognise the effect of a single 4-car failure when it is in service as part of a 12-car train.

 

[swt_passenger]

As an aside, is there an accessible online source of MTIN figures for all MUs? Many posters here quote them but maybe they are from sources available only to industry insiders. It would be useful to compare the oft quoted high figures of established three and four-car EMUs with the current growth of new 5/6/7/8/9 & 12-car units, particularly as the MTIN for example doesn't recognise the effect of a single 4-car failure when it is in service as part of a 12-car train.

No they aren't publically online, which is why people are usually quoting Modern Railways. Many such discussions have coincided with the January edition of the magazine, which allows most classes to be compared like you suggest.

 

[AM9]

No they aren't publically online, which is why people are usually quoting Modern Railways. Many such discussions have coincided with the January edition of the magazine, which allows most classes to be compared like you suggest.

Thanks for the reply. As they would be useful for general discussion of such matters, I wonder if a brief list could be posted here. The information I presume would be available as a FoI request and if so the raw figures would not be subject to copyright.

 

[samuelmorris]

The number of cars per unit also has a big effect and whether they are join /split in service.
e.g. 8car 700 should be better than 12car 700 for example.
Any computer reset will take more than 3 minutes from the start of the problem so that could be contributing a lot. Getting software sorted doesn't really fit in with traditional MTIN type thinking.
With the 700s you have to wonder whether the comparative lack of maintenance time and staff is having a negative self reinforcing effect outside any software issues?

707s - that is good door reliability to achieve that. And not coupling units in service will also help.

The 700s I would now describe as 'fair'. 14-15k for units roughly half of which are 8-car and half 12-car is comensurate with legacy EMUs with figures in the mid 20s. It's not great, but it's not dreadful either.
The 717s being sub-10k still have a lot further to go but they are demonstrating improvement having doubled in performance in the 6 months when no data was provided. They are also a lot less of a mature fleet. The 707s, however, are outstanding - not just the top new fleet, but the top fleet full stop. Unless there's some SWT-style creative accounting going on, they are definitely the new benchmark standard.

 

[hwl]

The 700s I would now describe as 'fair'. 14-15k for units roughly half of which are 8-car and half 12-car is comensurate with legacy EMUs with figures in the mid 20s. It's not great, but it's not dreadful either.
The 717s being sub-10k still have a lot further to go but they are demonstrating improvement having doubled in performance in the 6 months when no data was provided. They are also a lot less of a mature fleet. The 707s, however, are outstanding - not just the top new fleet, but the top fleet full stop. Unless there's some SWT-style creative accounting going on, they are definitely the new benchmark standard.

700s and 717s have AC-DC change on route which is another common failure mode that most units don't have. Worth having a look at the old FCC 377/5 performance vs other DC Electrostar performance (it was very noticeably worse).

 

[samuelmorris]

700s and 717s have AC-DC change on route which is another common failure mode that most units don't have. Worth having a look at the old FCC 377/5 performance vs other DC Electrostar performance (it was very noticeably worse).

Of course, but I don't imagine they were one thirteenth the rating, as 700s are versus 707s.

 

[hwl]

Of course, but I don't imagine they were one thirteenth the rating, as 700s are versus 707s.

I have feeling that it was about 2.5-3 :1 factor for dual voltage and then you have unit length so if you have 3 units making up a 12car the nominal per train rather than per units that is another multiplier factor.
Shorter units traditionally did better was there was less equipment per unit to fail but potentially more units per train.
So length and AC-DC issues could easily be between 30-70% of the difference.

With better reliability in general and much more varied unit lengths MTIN is potentially a less useful comparator than it once was.

 

[swt_passenger]

Thanks for the reply. As they would be useful for general discussion of such matters, I wonder if a brief list could be posted here. The information I presume would be available as a FoI request and if so the raw figures would not be subject to copyright.

I doubt you can FOI them, to the best of my knowledge they aren’t collated by a public body. I don’t think anyone’s going to copy a few pages of January’s Modern Railways to the forum either.

 

[samuelmorris]

I doubt you can FOI them, to the best of my knowledge they aren’t collated by a public body. I don’t think anyone’s going to copy a few pages of January’s Modern Railways to the forum either.

Yeah it's slightly questionable posting detailed prose about the figures, but posting the entire chart in the forum would really be breach of copyright. It's a shame they aren't collated by a public body but just how it is. If MRM are individually filing requests with the TOCs to get the data, they deserve to be remunerated for that work by people buying the magazine.

I have feeling that it was about 2.5-3 :1 factor for dual voltage and then you have unit length so if you have 3 units making up a 12car the nominal per train rather than per units that is another multiplier factor.
Shorter units traditionally did better was there was less equipment per unit to fail but potentially more units per train.
So length and AC-DC issues could easily be between 30-70% of the difference.

With better reliability in general and much more varied unit lengths MTIN is potentially a less useful comparator than it once was.

Skewing MTIN assessment based on unit length is something I've been told off for several times as rightly, there are some factors which there will be the same failure risk whether it's a 12-car or 4-car train, e.g. cab equipment. To that end I think it's probably fair to take somewhere in the middle. I'd equate an 8-car fixed-formation unit posting 10000 to a 4-car unit with more like 15000 rather than 20000. Meanwhile, even if we consider the 707s as an outlier, historically there are several classes that have all posted figures in the 90-120k range - 350/1, /2 and /3 subclasses, 360/1s, 379s, 444s and 450s all scored in that ballpark. That I feel is a reasonable expectation for solid, well understood and maintained EMUs. With the exception of the 444s, all of those are 4x20m units, so it is fair to accept lower for longer units, but not as much lower as we've yet seen.

 

[AM9]

I have feeling that it was about 2.5-3 :1 factor for dual voltage and then you have unit length so if you have 3 units making up a 12car the nominal per train rather than per units that is another multiplier factor.
Shorter units traditionally did better was there was less equipment per unit to fail but potentially more units per train.
So length and AC-DC issues could easily be between 30-70% of the difference.

With better reliability in general and much more varied unit lengths MTIN is potentially a less useful comparator than it once was.

It' interesting that you use the word "reliability" in this thread. I seem to remember that MTIN refers to something like 'miles to incident', (I'm sure somebody here will correct me if it isn't), which although would include reliability issues of certain items of hardware or subsystems, there are problems with trains that might be ascribed to causes other than a random failure and should be addressed through reliability measures by the design authority. You've mentioned issues regarding software reboots frequently causing a non-reliability issue to become an operating incident it terms of delays. Thus a comparison between new designs currently entering service vs. established stock where operating practice is probabbly more hohned towards dealing with their known quirks. One area might be door control where increased signalling/closing and subsequent interlock time alloyed with an increase of passengers forcing and holding doors open, has created incidents whereas with regular closing doors with no built-in checks would probably not breach whatever delay thresholds triggered an incident.
Feel free to correct me if I'm talking gibberish here.

 

[AM9]

Yeah it's slightly questionable posting detailed prose about the figures, but posting the entire chart in the forum would really be breach of copyright. It's a shame they aren't collated by a public body but just how it is. If MRM are individually filing requests with the TOCs to get the data, they deserve to be remunerated for that work by people buying the magazine.

OK, it's only through this thread that I've realised that the figures are published in one month's magazine each year and I don't have an issue buying it just for that reason. I'll keep my eyes open in December/January.

Skewing MTIN assessment based on unit length is something I've been told off for several times as rightly, there are some factors which there will be the same failure risk whether it's a 12-car or 4-car train, e.g. cab equipment. To that end I think it's probably fair to take somewhere in the middle. I'd equate an 8-car fixed-formation unit posting 10000 to a 4-car unit with more like 15000 rather than 20000. Meanwhile, even if we consider the 707s as an outlier, historically there are several classes that have all posted figures in the 90-120k range - 350/1, /2 and /3 subclasses, 360/1s, 379s, 444s and 450s all scored in that ballpark. That I feel is a reasonable expectation for solid, well understood and maintained EMUs. With the exception of the 444s, all of those are 4x20m units, so it is fair to accept lower for longer units, but not as much lower as we've yet seen.

Assuming my post #2475 above isn't completely gibberish, your words above party explain my gut feelings. With exceptions of the class 707s and the 379s all the quoted high scorers are getting on for 20 years old by design, the 379s are younger although much of their hardware is based on the original Electrostars. The 707s as hwl says are similar to 700s with some of the complicated bits removed which clearly favours their availability.

 

[hwl]

It' interesting that you use the word "reliability" in this thread. I seem to remember that MTIN refers to something like 'miles to incident', (I'm sure somebody here will correct me if it isn't), which although would include reliability issues of certain items of hardware or subsystems, there are problems with trains that might be ascribed to causes other than a random failure and should be addressed through reliability measures by the design authority. You've mentioned issues regarding software reboots frequently causing a non-reliability issue to become an operating incident it terms of delays. Thus a comparison between new designs currently entering service vs. established stock where operating practice is probably more hohned towards dealing with their known quirks. One area might be door control where increased signalling/closing and subsequent interlock time alloyed with an increase of passengers forcing and holding doors open, has created incidents whereas with regular closing doors with no built-in checks would probably not breach whatever delay thresholds triggered an incident.
Feel free to correct me if I'm talking gibberish here.

SWT were quite good at ascribing causes elsewhere!
The whole point is MTIN is good for trends within a class over time but less useful for comparison now as the underlying causes and trends aren't so transferable due to software and length variations.

The 700 are good example as their introduction aligned with radical operating changes whereas the 707s have fitted into the existing routines so there was a clear and certain specification. 700s are still a moving target in that respect with ATO roll out still to come. Hence taking along time to bed the 700s in inst a surprise.

With MTIN a small number of failure types define most of the failures e.g. doors issues / traction equipment on the networkers, eliminate those small number of issues and then the reliability stats improve as lots of items are incredibly reliable. In the 465 networker case the Brush ones were retractioned and are noticeably (+20%) more reliable than the GEC-Alstom ones. All of them got door reliability mods but that is still one of their biggest TIN causes

 

[jopsuk]

Thanks for the reply. As they would be useful for general discussion of such matters, I wonder if a brief list could be posted here. The information I presume would be available as a FoI request and if so the raw figures would not be subject to copyright.

No, TOCs aren't FOI-able (or where they are, this data would possibly be excluded). As far as I know Modern Railways/Roger Ford receives the information (through official channels) out of professional respect within the industry. The tables are therefore Copyright of Modern Railways.

 

[samuelmorris]

OK, it's only through this thread that I've realised that the figures are published in one month's magazine each year and I don't have an issue buying it just for that reason. I'll keep my eyes open in December/January.

For every fleet, yes. A small selection of 'new trains' covering most of the newcomers is published monthly. I'm not sure how far back that goes though, I only first subscribed this time last year.

 

[3141]

It' interesting that you use the word "reliability" in this thread. I seem to remember that MTIN refers to something like 'miles to incident'.

I believe MTIN stands for Miles per Technical Incident, and a technical incident is one that delays the train for more than three minutes. The number of such incidents does have a connection with reliability, though several incidents that had delayed each affected train for just over three minutes might look worse than just one that had delayed a single train for 90 minutes, with extensive knock-on delays to other services. It might then be argued that there should be a further measure recording the number of delay minutes each MTIN causes, but I don't feel it's worth compiling increasingly complex reports which become more and more difficult for people to grasp. The MTIN figures certainly reveal trends, and the individual TOCs will identify the issues which cause most incidents in each type of train.

 

[samuelmorris]

I believe MTIN stands for Miles per Technical Incident, and a technical incident is one that delays the train for more than three minutes. The number of such incidents does have a connection with reliability, though several incidents that had delayed each affected train for just over three minutes might look worse than just one that had delayed a single train for 90 minutes, with extensive knock-on delays to other services. It might then be argued that there should be a further measure recording the number of delay minutes each MTIN causes, but I don't feel it's worth compiling increasingly complex reports which become more and more difficult for people to grasp. The MTIN figures certainly reveal trends, and the individual TOCs will identify the issues which cause most incidents in each type of train.

A DPI (Delay per incident) measure is also recorded, for what it's worth.

 

[43096]

For every fleet, yes. A small selection of 'new trains' covering most of the newcomers is published monthly. I'm not sure how far back that goes though, I only first subscribed this time last year.

Modern Railways has been publishing the data annually since around 2004, I believe.

 

[Goldfish62]

I doubt you can FOI them, to the best of my knowledge they aren’t collated by a public body. I don’t think anyone’s going to copy a few pages of January’s Modern Railways to the forum either.

You can't FOI a private entity, irrespective of whether it's contracted to a public body. You can however FOI said public body in respect of stats / comms they might hold with the private entity, but if they refuse to disclose the information I'm not sure you'd get anywhere appealing

 

[PG]

Seems that the TOCs fleet reliability data is reported to ReFocus (Fleet Reliability Focus Forum) which was introduced by the Engineering Council 10 years ago.

See this document:
https://www.raildeliverygroup.com/files/Publications/engportal/TheTwentyPointPlan/20PPIssue13.pdf

I'm not sure if ReFocus, the Engineering Council or RDG would be FOI-able?

 

[Mordac]

You can't FOI a private entity, irrespective of whether it's contracted to a public body. You can however FOI said public body in respect of stats / comms they might hold with the private entity, but if they refuse to disclose the information I'm not sure you'd get anywhere appealing

Some private entities which receive public money are subject to FOI: universities for example. I'm not saying TOCs are, but it's not that simple.

 

[Goldfish62]

Some private entities which receive public money are subject to FOI: universities for example. I'm not saying TOCs are, but it's not that simple.

If, for example, someone submits an FOI request to a London bus operator it is not in scope and can be refused. However, if they send a request to TfL requesting information on that operator, eg email exchanges between TfL and the operator, then such information is in scope.

 

[Meerkat]

Even if you FOI a public entity they can refuse due to the need for commercial confidentiality.

 

[Goldfish62]

Even if you FOI a public entity they can refuse due to the need for commercial confidentiality.

If you refuse a whole request on that basis there has to be a sound justification. Simply redacting commercially sensitive content is the best way of avoiding a follow-up/appeal.

 

[AM9]

If you refuse a whole request on that basis there has to be a sound justification. Simply redacting commercially sensitive content is the best way of avoiding a follow-up/appeal.

And would the number of trust incidents on a class of rolling stock be commercially sensitive when every trust incident in service that affects the delivery of that service, (i.e. delays, cancellations, skip stopping etc.), is likely to be reported in the delay attribution process which is necessary to operate delay-repay and other compensation schemes? If of course the train dies in the stabling area and another is available to replace it such that the service runs as advertised, would that still influence MTIN figures?

 

[The Ham]

Could SWT/SWR be doing as simple as (say) doing a full software restart before the trains enter service? As if they've learnt that this then means that trains then tend to run fine for X hours or Y miles then this could be what's making the difference.

Likewise they may train their guards to do certain things first. For instance of a door problem is mainly fixed by doing one task they are told to do that first then try other things afterwards. That could allow the delay to be minimised and therefore not get counted as an incident.